Abstract:

Objective To develop a prediction model for identifying carbapenem-resistant Enterobacteriaceae （CRE） as the causative pathogen in bloodstream infections （BSI） using a DynNom dynamic nomogram， and to validate its predictive performance. Methods Patients with Enterobacteriaceae BSI admitted to the First Affiliated Hospital and the Fourth Affiliated Hospital of Soochow University between January 2021 and January 2024 were enrolled. Based on antimicrobial susceptibility， patients were classified into a CRE group （n = 173） and a carbapenem-susceptible Enterobacteriaceae （CSE） group （n = 102）. Univariate and multivariate logistic regression analyses were employed to identify independent risk factors for a CRE （vs. CSE） etiology among these patients. A DynNom dynamic nomogram prediction model was subsequently constructed based on these factors. Results Malignancy， indwelling central venous catheter， prior use of carbapenem antibiotics， higher Pitt bacteremia score， and an absolute neutrophil count （ANC） < 0.5 × 10?/L were identified as independent risk factors for CRE BSI （P < 0.05）. Internal validation of the model yielded a C-index of 0.812 （95% CI： 0.787 - 0.837）. The calibration curve closely approximated the ideal line. The area under the receiver operating characteristic （ROC） curve was 0.817 （95% CI： 0.788 - 0.846）. Decision curve analysis indicated a positive net benefit of the model across a threshold probability range of 23% to 100%. Conclusions Malignancy， indwelling central venous catheter， prior carbapenem use， higher Pitt bacteremia score， and ANC < 0.5 × 10?/L are independent risk factors for a CRE etiology in patients with Enterobacteriaceae BSI. The DynNom dynamic nomogram developed based on these factors demonstrates good predictive value for discriminating between CRE and non-CRE pathogens in this patient population.

Key words: carbapenem-resistant Enterobacteriaceae, bloodstream infection, risk factors, DynNom dynamic scores